1. Prescriptive analytics
Now that you've learned about descriptive, diagnostic, and predictive analytics, it is time to learn about the fourth type of analytics: prescriptive analytics.
2. Analytics overview
Prescriptive analytics builds on the results of predictive analytics. While predictive analytics allows us to identify potential outcomes and their probabilities, prescriptive analytics aims to take this further and take action on the predicted outcome.
As can be deduced from the name, with prescriptive analytics, we are trying to answer the question "What is the 'prescribed' or recommended course of action?" and, by doing so, select the best action given the outcome we want to achieve and taking into account related benefits and drawbacks.
3. Why use prescriptive analytics?
The primary purpose of prescriptive analytics is to help us decide what best to do. In doing so, it provides the following benefits: the ability to make informed, data-driven decisions, optimization of processes, and mitigation of risks associated with particular outcomes.
4. Common techniques
Common techniques used in prescriptive analytics include rule-based systems, which consists of generating a set of rules or decision logic to get the best outcome. These rules are generated by domain knowledge, a machine learning model or a combination of both.
In reinforcement learning, an algorithm learns to achieve a particular objective or optimize an outcome by receiving positive and negative feedback when running through a set of actions.
Scenario analysis and simulation consist of running through a set of pre-determined scenarios or simulating multiple outcomes to help select the decision that leads to the best outcome.
And lastly, the recommendation engine is one of the most well-known techniques. We'll discuss this in further detail on the next slide.
5. Recommendation engine
You've probably seen this a lot: when shopping online or watching a movie on Netflix, you're recommended what to buy or watch next.
Behind these recommendations is a recommendation engine.
It first looks at the products you've viewed or purchased in the past and uses this as data to predict what other products you would be interested in purchasing. Then it provides specific recommendations based on these predicted interests.
6. Case study: fraud detection
Suppose you work in a bank and transactions on a particular account have been flagged as possibly fraudulent. How should we proceed? Block the account? Call the customer to double-check the transactions. Have a fraud analyst do a manual check?
Each of these options comes with its risks and benefits. We want to avoid fraud losses, but we would like to avoid the negative impact of completely blocking a client's account.
It is also possible we have limited resources. For example, the fraud analyst can only check a limited number of cases daily. Prescriptive analytics can assist in these kinds of considerations and decisions.
7. Let's practice!
Now that you've learned about prescriptive analytics let's put that knowledge into practice.